-
Release v0.6.8-dev
Pre-releaseSome checks failedBuild and Test / build (push) Failing after 10sBuild Container / Build and Push Container (push) Failing after 10sBuild Container / Create Release (push) Has been skippedBuild Linux / build-linux (linux-amd64, false, x86_64-unknown-linux-musl) (push) Failing after 20sBuild Linux / build-linux (linux-arm64, true, aarch64-unknown-linux-gnu) (push) Failing after 24sBuild macOS / build-macos (push) Has been cancelledreleased this
2026-03-23 14:45:17 +00:00 | 27 commits to development since this release- fix: add /v1 to aibroker endpoint path (was getting 404 on /chat/completions)
- 5 models aligned, ffmpeg included, voice working
herodev.gent04.grid.tf → hero_zero:0.6.8-dev
Downloads
-
Source code (ZIP)
0 downloads
-
Source code (TAR.GZ)
0 downloads
-
Release v0.6.7-dev
Pre-releaseSome checks failedBuild and Test / build (push) Failing after 9sBuild Container / Build and Push Container (push) Failing after 10sBuild Container / Create Release (push) Has been skippedBuild Linux / build-linux (linux-amd64, false, x86_64-unknown-linux-musl) (push) Failing after 22sBuild Linux / build-linux (linux-arm64, true, aarch64-unknown-linux-gnu) (push) Failing after 21sBuild macOS / build-macos (push) Has been cancelledreleased this
2026-03-23 14:17:39 +00:00 | 28 commits to development since this releaseChanges
- Aligned aibroker + hero_agent on 5 models: claude-sonnet-4.5, claude-haiku-4.5, gpt-4o, gpt-4o-mini, llama-3.3-70b-versatile
- Added GROQ_API_KEY to aibroker for llama backend
- Added ffmpeg to Docker image for voice transcription
- Fixed agent model names to match aibroker
herodev.gent04.grid.tf → hero_zero:0.6.7-dev
Downloads
-
Source code (ZIP)
0 downloads
-
Source code (TAR.GZ)
0 downloads
-
Release v0.6.5-dev
Pre-releaseSome checks failedBuild and Test / build (push) Failing after 10sBuild Container / Build and Push Container (push) Failing after 9sBuild Container / Create Release (push) Has been skippedBuild Linux / build-linux (linux-amd64, false, x86_64-unknown-linux-musl) (push) Failing after 17sBuild Linux / build-linux (linux-arm64, true, aarch64-unknown-linux-gnu) (push) Failing after 22sBuild macOS / build-macos (push) Has been cancelledreleased this
2026-03-23 13:17:10 +00:00 | 31 commits to development since this releaseChanges since v0.6.4-dev
- fix: voice dropdown readable in dark mode (background-color from theme vars)
- fix: add ffmpeg to Docker image — required for voice transcription (WebM→MP3 conversion for Groq Whisper)
Note on ffmpeg
Both hero_agent and hero_books mic buttons require ffmpeg to convert browser-recorded WebM to MP3 before sending to Groq. This was missing from all previous container images.
Cumulative since v0.6.0-dev
- Issues #74 (Voice AI), #75 (Knowledge Pipeline), #76 (smoke fixes), #77 (UI polish) — all closed
- Phase 2 planned in #78 (cross-browser wake word + local Whisper)
- Full WASM shell, aibroker-first models, button contrast, semver deploy, ffmpeg
Running on
herodev.gent04.grid.tf → hero_zero:0.6.5-dev
Downloads
-
Source code (ZIP)
0 downloads
-
Source code (TAR.GZ)
0 downloads
-
Release v0.6.4-dev
Pre-releaseSome checks failedBuild and Test / build (push) Failing after 10sBuild Container / Build and Push Container (push) Failing after 9sBuild Container / Create Release (push) Has been skippedBuild Linux / build-linux (linux-amd64, false, x86_64-unknown-linux-musl) (push) Failing after 17sBuild Linux / build-linux (linux-arm64, true, aarch64-unknown-linux-gnu) (push) Failing after 22sBuild macOS / build-macos (push) Has been cancelledreleased this
2026-03-23 03:01:51 +00:00 | 31 commits to development since this releaseChanges since v0.6.1-dev
v0.6.4-dev — UI polish (#77)
- Voice buttons: consistent sizing, visible borders in OFF state, works in dark+light themes
- Auto-read toggle: added "Read" label, bigger 16px icon
- Model dropdown: only shows aibroker models (claude-sonnet-4.5, etc.), hides OpenRouter internals
v0.6.3-dev — hero_os_ui fix
- Fixed hero_os_ui/server service TOML: removed
servearg (broken by upstream CLI refactor e636a54) - Fixed model dropdown order: aibroker models first
v0.6.2-dev — WASM shell
- Full WASM rebuild included (hero_os_ui desktop shell loads)
- Fixed stale Docker pack (service TOMLs were missing)
Issues closed
- #77: Voice button contrast + hide OpenRouter models
- #76: Smoke test failures (Redis auth, password, libraries)
- #75: AI Agent knowledge pipeline
- #74: Voice AI enhancements
Deploy
TAG=0.6.4-dev make pack cd deploy/single-vm && make update ENV=herodev VERSION=0.6.4-devCurrently running
herodev.gent04.grid.tf → hero_zero:0.6.4-dev
Downloads
-
Source code (ZIP)
0 downloads
-
Source code (TAR.GZ)
0 downloads
-
Release v0.6.1-dev
Pre-releaseSome checks failedBuild and Test / build (push) Failing after 15sBuild Container / Build and Push Container (push) Failing after 10sBuild Container / Create Release (push) Has been skippedBuild Linux / build-linux (linux-arm64, true, aarch64-unknown-linux-gnu) (push) Failing after 19sBuild Linux / build-linux (linux-amd64, false, x86_64-unknown-linux-musl) (push) Failing after 23sBuild macOS / build-macos (push) Has been cancelledreleased this
2026-03-22 20:15:37 +00:00 | 32 commits to development since this releaseChanges (from v0.6.0-dev)
- fix: hero_redis auth.validate RPC handler for UI login (#76)
- fix: auth smoke test password (admin → admin123)
- fix: Redis smoke test uses /api/command instead of /rpc proxy (transport mismatch)
- fix: libraries.list graceful skip when repos still cloning
- fix: hero_os_ui tests skip when WASM not built
- fix: remove stale hero_shrimp_ui test
- feat: semver-based deployment (VERSION= in Makefile, version tracked in app.env)
- docs: mycelium-restart tip in CLAUDE.md
Deploy
TAG=0.6.1-dev make pack cd deploy/single-vm && make update ENV=herodev VERSION=0.6.1-devDownloads
-
Source code (ZIP)
0 downloads
-
Source code (TAR.GZ)
0 downloads
-
Release v0.6.0-dev
Pre-releaseSome checks failedBuild and Test / build (push) Failing after 12sBuild Container / Build and Push Container (push) Failing after 13sBuild Container / Create Release (push) Has been skippedBuild Linux / build-linux (linux-amd64, false, x86_64-unknown-linux-musl) (push) Failing after 16sBuild Linux / build-linux (linux-arm64, true, aarch64-unknown-linux-gnu) (push) Failing after 28sBuild macOS / build-macos (push) Has been cancelledreleased this
2026-03-22 17:48:14 +00:00 | 35 commits to development since this releaseChanges
- feat: hero_agent ecosystem integration — MCP wiring, LLM routing, system prompt (#75)
- feat: search_hero_docs tool, OpenRPC spec, hero_proc_sdk lifecycle (#75)
- feat: skills tab in admin dashboard with execution statistics (#75)
- feat: voice AI — auto-read, voice selector, conversation mode, wake word (#74)
- fix: aibroker daemon broken hero_proc_sdk types (#75)
- smoke tests for hero_agent (health, OpenRPC, config, skills, voice)
Deploy
TAG=0.6.0-dev make pack && make push cd deploy/single-vm && make update ENV=herodevDownloads
-
Source code (ZIP)
0 downloads
-
Source code (TAR.GZ)
0 downloads
-
Release v0.5.0-dev
Pre-releaseSome checks failedBuild and Test / build (push) Failing after 12sBuild Container / Create Release (push) Has been skippedBuild Container / Build and Push Container (push) Failing after 9sBuild Linux / build-linux (linux-amd64, false, x86_64-unknown-linux-musl) (push) Failing after 15sBuild Linux / build-linux (linux-arm64, true, aarch64-unknown-linux-gnu) (push) Failing after 29sBuild macOS / build-macos (push) Has been cancelledreleased this
2026-03-22 16:45:14 +00:00 | 36 commits to development since this releaseHero Services v0.5.0-dev
Major release: hero_agent replaces hero_shrimp, voice AI, SSE streaming.
New Features
- hero_agent replaces hero_shrimp — Rust-native AI agent with 57 built-in tools, SSE streaming, MCP integration (#72)
- Voice input in AI Assistant — mic button, herolib_ai STT (Groq Whisper + ffmpeg), auto-transcribe and send (#73, #74)
- Read aloud — speaker button on AI responses, server TTS with browser Speech API fallback (#74)
- AI Agent admin dashboard — Voice tab with 3 modes (push-to-talk, VAD, conversation) (#74)
- SSE streaming — word-by-word AI responses in the AI Assistant (#32)
- Voice chat API —
/api/voice/chat(STT→Agent→TTS),/api/voice/transcribe,/api/voice/tts(#73)
Fixes
- Connection status indicator fixed for Voice + Indexer UIs (#70)
- Proxy root redirects to Hero OS when
HERO_PROXY_DEFAULT_SERVICEis set (#70) - Socat bridge fixed: port 6666 → proxy server TCP:9997 (#70)
- Docker overlay2 on ext4 /data (no fuse-overlayfs needed) (#71)
- hero_services_server CLI args updated for upstream refactor
- TTS model updated to gpt-4o-mini-tts (12x cheaper)
Issues Closed
- #70 Connection status indicator
- #71 ext4 overlay2
- #72 hero_agent integration
- #73 Voice AI backend
- #32 SSE streaming
Repos touched
hero_agent, hero_services, hero_archipelagos, hero_os, hero_voice, hero_indexer_ui, hero_proxy
Deploy
TAG=herodev make pack && make push cd deploy/single-vm && make update ENV=herodevLive at https://herodev.gent04.grid.tf
Downloads
-
Source code (ZIP)
0 downloads
-
Source code (TAR.GZ)
0 downloads
-
Release v0.4.0-dev
Pre-releaseSome checks failedBuild and Test / build (push) Failing after 14sBuild Container / Build and Push Container (push) Failing after 13sBuild Container / Create Release (push) Has been skippedBuild Linux / build-linux (linux-amd64, false, x86_64-unknown-linux-musl) (push) Failing after 18sBuild Linux / build-linux (linux-arm64, true, aarch64-unknown-linux-gnu) (push) Failing after 20sBuild macOS / build-macos (push) Has been cancelledreleased this
2026-03-21 06:21:28 +00:00 | 47 commits to development since this releaseHero Services v0.4.0-dev
Connection status indicator fixes and proxy routing improvements.
Changes
Connection status indicator (#70 follow-up)
- hero_voice_ui: Moved status dot from bottom status bar to navbar header next to "HeroVoice" brand. Popover now opens downward.
- hero_indexer_ui: Added clickable popover to the Askama-templated dashboard (the actual served UI at
lhumina_code/hero_indexer_ui/, not the rust-embed static file).
Proxy routing fixes
- hero_proxy_server: Root handler (
/) now redirects to default service whenHERO_PROXY_DEFAULT_SERVICEenv var is set. Service discovery dashboard still shown when unset. - entrypoint.sh: Fixed socat bridge — was routing port 6666 to
hero_proxy_ui.sock(admin dashboard only), now routes tohero_proxy_serverTCP:9997 (the actual reverse proxy). - hero_proxy.toml: Added
HERO_PROXY_DEFAULT_SERVICE=hero_os_uiso navigating to root shows Hero OS.
UI pattern reference
Pattern Services Shared JS module ( connection-status.js)proc, books, embedder, inspector, browser, collab, whiteboard, foundry Inline JS poller (rust-embed) voice, proxy, auth Askama template + inline JS indexer, osis, biz Smoke tests
111 passed, 0 failed, 2 skipped.
Repos touched
hero_voice— voice UI HTMLhero_indexer_ui— Askama base templatehero_proxy— proxy server root redirecthero_services— entrypoint + proxy config
Signed-off-by: mik-tf
Downloads
-
Source code (ZIP)
0 downloads
-
Source code (TAR.GZ)
0 downloads
-
Release v0.1.0
Stablereleased this
2026-02-28 12:51:14 +00:00 | 200 commits to development since this releaseHero Services Container Release v0.1.0
Production container with all service binaries pre-built.
Pull and run:
docker pull forge.ourworld.tf/lhumina_code/hero_zero:0.1.0 docker run --rm -it -p 8804:8804 -p 3875:3875 forge.ourworld.tf/lhumina_code/hero_zero:0.1.0Downloads
-
Source code (ZIP)
0 downloads
-
Source code (TAR.GZ)
0 downloads
-
Source code (ZIP)